The AI agent platform for people who think busywork is boring - built in plain English, wired to your real data.
Cotera, New York. A banner for a company whose whole argument is that the interesting part of AI isn't the model - it's the work that quietly gets done after you close the laptop.
There is a certain genre of enterprise software pitch that promises to change everything, reinvent the way you work, and possibly your sense of self. Cotera, a roughly eleven-person startup headquartered in New York, has a pitch that is refreshingly smaller and, if you have ever worked in operations, considerably more honest: it is, per its own website, "for people who think busywork is boring." That is the whole idea. Not transformation. Just the tedious part, done.
Here is the mechanism. Cotera lets a person - specifically a go-to-market or operations person, not necessarily an engineer - build an AI agent by describing what they want in plain English. No flowchart. No code. You write something like you would write a brief for a new coworker, and the agent goes and does it: routes the support ticket, pulls the order status, processes the refund, enriches the lead, reads the customer feedback and tells you what people actually said. Then it writes the result back into the tools you already use, on a schedule, without being asked again.
The companies that win the next decade won't be the ones who buy the most models. They'll be the ones whose agents quietly do real work.
The interesting bet buried in that sentence is contrarian in a market that is not currently rewarding contrarianism. Everyone is buying models. The models are, by most accounts, extraordinary. And yet the gap between "we have access to a frontier model" and "a specific piece of work reliably got done" remains stubbornly wide, and it is mostly a plumbing problem. An agent that cannot see your real numbers is just a confident guesser. So Cotera's first move is unglamorous: it connects agents directly to the data warehouse - Snowflake, BigQuery, Postgres - and to the CRM, the support desk, the marketing stack. Two hundred-plus integrations, hand-built rather than assembled from a generic connector kit. The company is somewhat proud of not doing the "MCP kitchen-sink."
That choice tells you something about the taste of the founders. Ibby Syed, the co-founder and chief executive, spent years as a lead data scientist and manager of data science at Peloton before starting the company - which is to say he spent years watching capable people spend their days on work that a machine could plausibly do, if only the machine could reach the data. His co-founder, Tom Firth, rounds out a team split between New York and London. Peloton, in a tidy bit of narrative symmetry, later became a customer.
The design philosophy has a tagline - "Smarter Than Workflows. Easier Than Code." - and it is doing real work. Flowcharts, the old way, made software reliable but rigid. Raw prompting made it flexible but unpredictable. Cotera's claim is that you should not have to pick: write a document, get a dependable agent, keep fine-grained control over what it is allowed to touch. Whether that middle path holds up under enterprise load is the kind of question that gets answered slowly, one customer at a time. But the customer list is not nothing.
Bilt, Peloton, SKIMS, OpenSea, Coterie, Olipop, Shippo - a spread across commerce, crypto, and consumer that suggests the appeal is not confined to one vertical. And notably, the people building the agents at these companies are frequently not engineers. They are the ops and support and success teams, the ones closest to the busywork, now authoring their own automation. That shift - automation written by the people who feel the pain rather than by a backlog-bound engineering team - is arguably the more durable story here than any individual feature.
Cotera went through Y Combinator's Winter 2022 batch and reached general availability of the platform in December 2025, which in startup terms means the long, quiet stretch of building the thing is over and the louder stretch of selling it has begun. The funding is modest by the standards of a space where nine figures is now table stakes for a slide deck. That may be the point. If the thesis is that work done beats models bought, you would expect the company proving it to be spending its energy on the work, not the round.
Write the task in plain English. The agent deploys on a schedule and takes real action - no flowcharts, no code.
CRMs, support desks, marketing and ad tools, warehouses (Snowflake, BigQuery, Postgres) and any HTTP API.
Agents pull from your systems and push results back in, closing the loop instead of dumping into a dashboard.
A human-friendly interface to build, test and constrain agents - so you always know what they can touch.
Built-in assistance to sharpen your agent instructions when plain English needs a nudge.
Ticket routing, voice-of-customer analysis, lead enrichment, account-health monitoring, refund processing.
Cotera's agents read from and write to the tools teams already run. A sample of the 200+ integrations:
Former Lead Data Scientist and Manager of Data Science at Peloton; Entrepreneur in Residence at Steadview Capital before Cotera.
Co-founder of Cotera (YC W22), building the agent platform across the company's New York and London teams.
Product demos & talks
Its marketing openly says Cotera is "for people who think busywork is boring" - a rare bit of anti-hype in enterprise AI.
Rather than generic connectors, it favors 200+ hand-built integrations for reliability.
Co-founder Ibby Syed led data science at Peloton - which later became a Cotera customer.
The small team is split between New York City and London.
You build an agent the way you'd brief a coworker - in plain English, no flowchart required.
"Build an AI agent by writing a document. No flowcharts, no code - just plain English that takes real action on real data."
"The winners of the next decade won't buy the most models - they'll have agents quietly doing real work."
"Bilt, Peloton, SKIMS and OpenSea are building agents on Cotera - connected to the warehouse, not a chatbot demo."
"Cotera: smarter than workflows, easier than code. AI agents that actually know your business."
Cotera is a New York-based, Y Combinator-backed AI platform that lets go-to-market and operations teams build autonomous agents by describing tasks in plain English - no flowcharts and no code. The agents connect directly to a company's data warehouse, CRM, and 200+ business tools, then take real action on real data: routing support tickets, analyzing customer feedback, enriching leads, monitoring account health, and automating repetitive back-office work. Founded in 2021 by Ibby Syed and Tom Firth, Cotera pitches itself for people who think busywork is boring.
Last updated: